A Novel Sea Ice Classification Method from Hyperspectral Image Based on Bagging PCA Hashing

Hyperspectral imagery has evident advantage for sea ice classification due to enormous spectral bands. At the meantime, hashing is promising in representing a high-dimensional feature with extremely low bit binary codes while maintaining the classification performance. In this paper, a novel sea ice classification framework based on PCA Bagging hashing is proposed to improve the accuracy of hyperspectral image classification. First, local binary pattern(LBP) features are extracted from selected spectral bands. PCA bagging is employed to generating hashing codes of the extracted features. Then, a piece of short but strong code is obtained to represent the features of each pixel. Finally, these hashing codes are fed into the extreme learning machine for classification. The experimental results in real-world dataset demonstrate that the proposed framework is superior to two closely related methods.